Managing massive MIMO antennas in a wireless network

Abstract

Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a network entity may calculate, using a machine learning (ML) model trained to estimate an impact of a reconfiguration of an antenna on a set of key performance indicators (KPIs) of a given cell and one or more neighbors of the given cell, one or more predicted KPIs using data characterizing a reconfiguration of a massive multiple-input multiple-output (M-MIMO) antenna. The network entity may provide the one or more predicted KPIs. Numerous other aspects are described.

What is claimed is: 
     
         1 . An apparatus, comprising:
 a memory; and   one or more processors, coupled to the memory, configured to:
 calculate, using a machine learning (ML) model trained to estimate an impact of a reconfiguration of an antenna on a set of key performance indicators (KPIs) of a given cell and one or more neighbors of the given cell, one or more predicted KPIs using data characterizing a reconfiguration of a massive multiple-input multiple-output (M-MIMO) antenna, wherein the ML model comprises:
 a first input ML sub-model configured as a feed forward artificial neural network configured to process a set of first input data; 
 a second input ML sub-model configured as a natural language processing neural network configured to process a set of second input data; and 
 a third ML sub-model configured as a feed forward artificial neural network configured to aggregate an output of the first input ML sub-model and an output of the second input ML sub-model; and 
 
 provide the one or more predicted KPIs for update of the M-MIMO antenna based on the ML model. 
   
     
     
         2 . The apparatus of  claim 1 ,
 wherein the data characterizing the reconfiguration of the M-MIMO antenna includes the set of first input data characterizing M-MIMO configuration parameters and the set of second input data related to one or more transitions between M-MIMO configuration states.   
     
     
         3 . The apparatus of  claim 2 ,
 wherein the output of the first input ML sub-model includes information regarding a radio link environment and the output of the second input ML sub-model includes information regarding a preceding configuration transition and one or more next states of the M-MIMO antenna.   
     
     
         4 . The apparatus of  claim 1 ,
 wherein the first input ML sub-model and the second input ML sub-model are configured to process the set of first input data and the set of second input data in parallel.   
     
     
         5 . The apparatus of  claim 2 ,
 wherein the set of second input data comprises a natural language processing based representation of the one or more transitions.   
     
     
         6 . The apparatus of  claim 1 ,
 wherein the one or more predicted KPIs include at least one of:
 a coverage parameter, 
 a load parameter, 
 a throughput parameter, 
 an interference parameter, 
 an inter-beam handoff parameter, 
 an inter-site handoff parameter, or 
 a combination thereof. 
   
     
     
         7 . The apparatus of  claim 1 ,
 wherein the one or more processors are further configured to train the ML model.   
     
     
         8 . The apparatus of  claim 7 , wherein the one or more processors, to train the ML model, are configured to train the first input ML sub-model, the second input ML sub-model, and the third ML sub-model of the ML model jointly. 
     
     
         9 . The apparatus of  claim 1 ,
 wherein the one or more processors, to provide the one or more predicted KPIs, are configured to provide the one or more predicted KPIs as an input to an algorithm to optimize a KPI based criterion of a group of cells.   
     
     
         10 . A method of managing an antenna configuration in a wireless network, comprising:
 calculating, using a machine learning (ML) model trained to estimate an impact of a reconfiguration of an antenna on a set of key performance indicators (KPIs) of a given cell and one or more neighbors of the given cell, one or more predicted KPIs using data characterizing a reconfiguration of a massive multiple-input multiple-output (M-MIMO) antenna, wherein the ML model comprises:
 a first input ML sub-model configured as a feed forward artificial neural network configured to process a set of first input data; 
 a second input ML sub-model configured as a natural language processing neural network configured to process a set of second input data; and 
 a third ML sub-model configured as a feed forward artificial neural network configured to aggregate an output of the first input ML sub-model and an output of the second input ML sub-model; and 
   providing the one or more predicted KPIs for updating the M-MIMO antenna based on the ML model.   
     
     
         11 . The method of  claim 10 ,
 wherein the data characterizing the reconfiguration of the M-MIMO antenna includes the set of first input data characterizing M-MIMO configuration parameters and the set of second input data related to one or more transitions between M-MIMO configuration states.   
     
     
         12 . The method of  claim 11 ,
 wherein the output of the first input ML sub-model includes information regarding a radio link environment and the output of the second input ML sub-model includes information regarding a preceding configuration transition and one or more next states of the M-MIMO antenna.   
     
     
         13 . The method of  claim 10 ,
 wherein the first input ML sub-model and the second input ML sub-model are configured to process the set of first input data and the set of second input data in parallel.   
     
     
         14 . The method of  claim 11 ,
 wherein the set of second input data comprises a natural language processing based representation of the one or more transitions.   
     
     
         15 . The method of  claim 10 ,
 wherein the one or more predicted KPIs include at least one of:
 a coverage parameter, 
 a load parameter, 
 a throughput parameter, 
 an interference parameter, 
 an inter-beam handoff parameter, 
 an inter-site handoff parameter, or 
 a combination thereof. 
   
     
     
         16 . The method of  claim 10 , further comprising training the ML model. 
     
     
         17 . The method of  claim 16 , wherein training the ML model further comprises training at least the first input ML sub-model and the second input ML sub-model jointly. 
     
     
         18 . The method of  claim 10 ,
 wherein providing the one or more predicted KPIs further comprises providing the one or more predicted KPIs as an input to an algorithm to optimize a KPI based criterion of a group of cells.   
     
     
         19 . A non-transitory computer-readable medium storing a set of instructions for wireless communication, the set of instructions comprising:
 one or more instructions that, when executed by one or more processors of an apparatus, cause the apparatus to:
 calculate, using a machine learning (ML) model trained to estimate an impact of a reconfiguration of an antenna on a set of key performance indicators (KPIs) of a given cell and one or more neighbors of the given cell, one or more predicted KPIs using data characterizing a reconfiguration of a massive multiple-input multiple-output (M-MIMO) antenna, wherein the ML model comprises: 
 a first input ML sub-model configured as a feed forward artificial neural network configured to process a set of first input data; 
 a second input ML sub-model configured as a natural language processing neural network configured to process a set of second input data; and 
 a third ML sub-model configured as a feed forward artificial neural network configured to aggregate an output of the first input ML sub-model and an output of the second input ML sub-model; and 
   provide the one or more predicted KPIs for update of the M-MIMO antenna based on the ML model.   
     
     
         20 . The non-transitory computer-readable medium of  claim 19 ,
 wherein the data characterizing the reconfiguration of the M-MIMO antenna includes the set of first input data characterizing M-MIMO configuration parameters and the set of second input data related to one or more transitions between M-MIMO configuration states.   
     
     
         21 . The non-transitory computer-readable medium of  claim 20 ,
 wherein the output of the first input ML sub-model includes information regarding a radio link environment and the output of the second input ML sub-model includes information regarding a preceding configuration transition and one or more next states of the M-MIMO antenna.   
     
     
         22 . The non-transitory computer-readable medium of  claim 20 ,
 wherein the set of second input data comprises a natural language processing based representation of the one or more transitions.   
     
     
         23 . The non-transitory computer-readable medium of  claim 19 ,
 wherein the first input ML sub-model and the second input ML sub-model are configured to process the set of first input data and the set of second input data in parallel.   
     
     
         24 . The non-transitory computer-readable medium of  claim 19 ,
 wherein the one or more predicted KPIs include at least one of:
 a coverage parameter, 
 a load parameter, 
 a throughput parameter, 
 an interference parameter, 
 an inter-beam handoff parameter, 
 an inter-site handoff parameter, or 
 a combination thereof. 
   
     
     
         25 . An apparatus for wireless communication, comprising:
 means for calculating, using a machine learning (ML) model trained to estimate an impact of a reconfiguration of an antenna on a set of key performance indicators (KPIs) of a given cell and one or more neighbors of the given cell, one or more predicted KPIs using data characterizing a reconfiguration of a massive multiple-input multiple-output (M-MIMO) antenna, wherein the ML model comprises:
 a first input ML sub-model configured as a feed forward artificial neural network configured to process a set of first input data; 
 a second input ML sub-model configured as a natural language processing neural network configured to process a set of second input data; and 
 a third ML sub-model configured as a feed forward artificial neural network configured to aggregate an output of the first input ML sub-model and an output of the second input ML sub-model; and 
   means for providing the one or more predicted KPIs for updating the M-MIMO antenna based on the ML model.   
     
     
         26 . The apparatus of  claim 25 ,
 wherein the data characterizing the reconfiguration of the M-MIMO antenna includes the set of first input data characterizing M-MIMO configuration parameters and the set of second input data related to one or more transitions between M-MIMO configuration states.   
     
     
         27 . The apparatus of  claim 26 ,
 wherein the output of the first input ML sub-model includes information regarding a radio link environment and the output of the second input ML sub-model includes information regarding a preceding configuration transition and one or more next states of the M-MIMO antenna.   
     
     
         28 . The apparatus of  claim 26 ,
 wherein the set of second input data comprises a natural language processing based representation of the one or more transitions.   
     
     
         29 . The apparatus of  claim 25 ,
 wherein the first input ML sub-model and the second input ML sub-model are configured to process the set of first input data and the set of second input data in parallel.   
     
     
         30 . The apparatus of  claim 25 ,
 wherein the one or more predicted KPIs include at least one of:
 a coverage parameter, 
 a load parameter, 
 a throughput parameter, 
 an interference parameter, 
 an inter-beam handoff parameter, 
 an inter-site handoff parameter, or 
 a combination thereof.